Improving Social Relationships in Child Care through a Researcher-Program Partnership

Early Education & DevelopmentVolume 15, Number 1, January 2004

Improving Social Relationships in Child Carethrough a Researcher-Program Partnership

Carollee Howes

Eva M. Shivers

Sharon Ritchie

University of California at Los Angeles

We describe the process of relationship-based intervention and associated changesin children and teachers’ relationships and behaviors within center-based childcare and examine the circumstances under which changes were most likely tooccur. The relationship-based intervention had three components: research partnersin each participating classroom, focus groups, and feedback sessions. Twenty-eight teachers and seventy children from 10 programs participated in the research.Approximately half of the children and one-third of the teachers were Latino, andalmost half of the teachers and one-third of the children African-American. Weused five distinct procedures: (1) participant observation; (2) naturalisticobservations using a time sampling procedure; (3) global ratings; (4) clinicalinterviews and (5) case studies. There were significant main effects for changeover time in children’s attachment security, complexity of play with peers, andtheir experiences of teacher responsive involvement. Teachers with BA degreeswere more engaged in the intervention than teachers with less formal education.When we held BA degrees constant, children who had teachers engaged in theintervention experienced the greatest changes in classroom climate, teacherresponsivity, and in their relationships with the teacher.

This work was supported by grants to the National Center for Early Development and Learning under theEducational Research and Development Centers Program, PR/Award Number R307A60004, as adminis-tered by the Office of Educational Research and Improvement, US Dept. of Education. We wish to ac-knowledge the participation and time of the programs, teachers, and children who allowed us to record theiractions, words, and practices. Finally, we would like to acknowledge our colleagues’ insights as we devel-oped this research. Thanks in particular to Allison Fuligini’s comments on an earlier draft of the manu-script. Correspondence concerning this article should be addressed to Carollee Howes, Graduate School ofEducation & Information Studies, University of California, Los Angeles, CA 90095-1521. Electronic mailmay be sent via Internet to [email protected].

151Howes.p65 11/25/03, 10:30 AM57

58 Howes, Shivers, & Ritchie

Improving Social Relationships in Child Care through aResearcher-Program Partnership

Whether an individual child’s experiences in child care enhance or compromise his/hersocial and emotional development depends in large part on the relationships the childconstructs with adults and children within the child care setting (Shonkoff & Phillips, 2000).Children and teachers have the opportunity to construct relationships that are independentof children’s prior adult-child relationships (Howes & Hamilton, 1992). However, bothteachers and children vary in their ability to construct positive relationships. Teachers whosimply follow the lead of the child tend to form positive relationships with children who aresocially competent, and negative relationships with children who have difficulties withrelationships. Children who have had difficult relationships with adults before entering childcare present a particular dilemma to their teachers because they tend to act as if the adult cannot be trusted, setting the stage for replicating their prior difficult social relationships (Howes& Ritchie 2002; Sroufe, 1983). Some teachers participate with these difficult children inreplicating negative adult-child relationships while other teachers manage to help themconstruct positive adult-child relationships (Denham & Burton, 1996; (Howes & Ritchie2002; Sroufe, 1983). The teachers who are able to change adult-child relationship qualitytowards more positive relationships do this by disconfirming children’s behaviors (Howes& Ritchie, 2002)

Problematic teacher-child relationships are not only disadvantageous for children inrelation to adults but as well for peer relations. Children who are preoccupied with contentiousadult-child relationships cannot construct positive relationships with peers (NICHD EarlyChild Care research Network, 2001). Children’s peer relationships, positive and problematic,become associated with their peers’ positive and negative expectations of them (Rubin,Bulowski, & Parker, 1998), and are therefore resistant to change once maladaptive patternsare established.

In this research, we attempted to intervene in the process of relationship formation inchild care via a relationship-based intervention. In relationship-based intervention a caregiveris supported as she/he works to remain sensitive and responsive to the children in their careeven when the children provoke anger or neglect (Erickson & Kurz-Riemer 1999; Lieberman& Zeanah, 1999). An extensive review and meta-analysis of relationship-based interventionprograms aimed at mothers, suggests that with relatively modest intervention adults can behelped to become more sensitive and responsive, and that as a result the relationships thatthey construct with children are more trusting (Van Ijzendoorn, Juffer, & Duyvesteyn, 1995).There is not a large corresponding literature to support such an intervention with adults whocare for children, but are not their mothers. However, when Denham and Burton (1996)provided child care teachers with one day of classroom training on relationship- buildingactivities and three days of classroom training on activities to assist children in understandingemotions and cognitive problem solving and on individualizing activities, the children inthese teachers’ classrooms had higher scores on social-emotional assessment than childrenin control group classrooms.

The first goal of this paper is to describe the process of intervention and changes inchildren and teachers’ relationships and behaviors. Our intervention was based onconstructing positive relationships between the researchers and the teachers. Thus, according

151Howes.p65 11/25/03, 10:30 AM58

59Improving Social Relationships in Child Care

to the principles of relationship-based intervention we expected to see changes in a narrowband of child care quality and children’s outcomes. Specifically we examined the emotionalclimate of the classroom, the responsive involvement of the teacher with particular children,and the children’s attachment security and social competence with peers. The emotionalclimate of the classroom is a relatively recent addition to the construct of child care quality.It refers to the tone of the classroom and can range from positive and prosocial to acrimoniousand harsh. Social –emotional classroom climates, in one study, predicted social competencewith peers in second grade (Howes, 2000). Responsive involvement describes interactionbetween adults and particular children. Responsive involvement can range fromindividualized and positive responsivity to non-responsive. In prior work, teacher responsiveinvolvement was associated with the security of child-teacher attachment (Howes & Hamilton,1992). We assumed that if the emotional climate of the classroom and the responsiveness ofthe teachers became more positive, there would also be positive changes in teacher-childattachment security and children’s social competence with peers.

The second related goal of this analysis was to examine predictors of change in socialrelationship quality: under what circumstances did the classroom emotional climate, teacherinvolvement, attachment security, and peer social competence, become more positive? Weidentified three predictors of positive change in these dimensions:teacher engagement in theintervention; teacher level of formal education in a child related field, and teacher-articulatedpractices around children’s socialization.

We could find no research literature linking the teacher’s level of engagement with theintervention and change in relationships and behavior. However, in studies of relationship-based intervention with mothers, when the mothers are more engaged with the intervenerthey are more likely to change (Erickson & Kurz-Riemer 1999). Since we assumed thatrelationship-based intervention with teachers would be similar in process to relationship-based intervention with mothers, we expected more engaged teachers to be associated withgreater changes.

We included teachers’ formal education in a child related field as a predictor because ofthe association between teacher formal education in a child related field and teacher responsiveinvolvement and effective teaching (Honig & Hirallal, 1998; Howes, James, & Ritchie,2003). Course work in a child related field is, in some research, considered independent offormal education, and in some studies more predictive of teacher behavior than simple yearsof education. In our current sample years of education and child related coursework wereconfounded. Most of the teachers received their post-secondary education as in-service ratherthan pre-service education (Howes et al, 2003) and none of the teachers with a BA degreehad a degree in a non-child related field. The modal teacher with a BA degree had taken herchild- related courses before her general education requirements. We reasoned that teacherswith BA degrees who had added general education to their child-related coursework mightbe less intimidated by the educational achievement of the graduate students who served asresearch partners, and thus find it easier to form positive relationships with them.

Teachers’ articulated practices are another relatively recent construct in this area ofresearch. In introducing the term “articulated practices,” we are using a definition of practicesrooted in sociocultural theory (Rogoff, 2002). Within this theoretical traditions, practicesare “ways of doing things” or ingrained habits rooted in participants beliefs, expectations,

151Howes.p65 11/25/03, 10:30 AM59


traditions, and relations. Teacher articulated practices refer to what teachers intend to teachas well as how they intend to teach. They are intentional strategies used to achieve an adultgoal, whether the goal is to have children be in quiet lines going to the bathroom, to have allthe children recognize ten letters, or to bolster self confidence. In our prior work, we foundhow teachers think and talk about what they do in the classroom to be related to ourobservations of what they and children actually do in classrooms (Wishard, Shivers, Howes,& Ritchie, 2003). In this work, we used teacher interviews about their practices as part ofthe intervention. Our intention was to begin to build teacher-researcher relationships bydemonstrating an interest in teacher’s thoughts about their work through careful listening ofthe teachers’ descriptions of what they meant to do in their classrooms. We predicted thatteachers who spontaneously told us that they worked with children to build positive socialrelationships would be associated with positive change.

We constrained our study in several important ways. The early childhood programsaccepted into the intervention were nominated by community advocates and by earlychildhood programs already participating in a long-term project exploring quality and practicesof child care for low-income, urban children of color. The criteria for program participationin this intervention project included having similar child and family demographics to theoriginal participating programs, and both a willingness and interest in engaging in thinkingand talking about their practices with researchers. Participation in the intervention was limitedto the newly nominated programs (See Wishard et al., 2003).

From the nominated programs we only selected programs that had average total itemscores of 4.0 or above on the Early Childhood Environmental Rating Scale in order to assurethat the program had sufficient material resources to focus on children’s social competence.Most of the children and teachers in the selected programs were Latino (primarily Mexicanor Central American immigrants) or African-American. Most of the families served by theprograms lived at or below the poverty line and were receiving subsidized care. Eight of theten programs participating in the intervention were involved in the intervention during thesame time period. The remaining programs were nominated at the same time as the others,but changes in location and staffing of the program resulted in a delay of several monthsbefore the intervention was begun. Therefore the staff in these programs were unable toparticipate in the focus group described below.

The Relationship-Based Intervention

The intervention lasted a child care school year—from September to July. Therelationship-based intervention had three components: research partners in each participatingclassroom, focus groups, and feedback sessions. Each classroom was assigned a graduatestudent research partner who was similar in ethnic background to the modal ethnic grouprepresented by program teaching staff and children. In all programs that used both Spanishand English in the classroom, the research partner was Spanish-English bilingual. The researchpartner visited the program once weekly. During all the visits, the partners engaged in thelife of the program, visiting classrooms, working with children under the direction of theteacher, and discussing events with the teachers during naps and breaks.

The relationship-based intervention component involving the research partner had severalsequential phases. The research partner spent the first weeks of her time in the program

151Howes.p65 11/25/03, 10:30 AM60


constructing a positive relationship with the teachers by orienting them to the project, beingwarm and responsive to the teachers as well as to the children, expressing positive interest inhow the teacher acted in the classroom and by answering the teachers’ questions and concernsabout participant observation. In the second phase, beginning after six visits and occurringsimultaneously with on-going visits, the research partner conducted individual interviewswith the teachers. These interviews took place when the teachers were not caring for childrenat a place and time the teachers selected. The interview was conducted in a conversationalformat and asked about teachers’ histories with children and teacher perceptions of theirown teaching. In the third phase, after the teachers had experienced feedback sessions (seebelow), the research partner initiated conversations with the teachers about the particularissues raised in the feedback sessions. During the entire project the research partnersparticipated in both group and individual supervision around their roles. The supervisors,one a licensed psychologist and the other an experienced teacher educator, provided theresearch partners with support and guidance.

The focus group component consisted of monthly focus groups comprised of two teacherrepresentatives from each program. The research partner and the project director met withthe teachers in each program and explained that the purpose of the focus groups was forteachers to talk and think about how they taught children. Teachers were paid to attend anddinner was provided. In all programs the teachers themselves, not the program director,decided who was to attend the focus groups. In all cases the research partner checked withthe teacher representative just before each focus group to make sure that she was able to attend.

A member of the research team who was not a research partner facilitated the focusgroups. Because she was not a research partner she was not allied with any particular programand she could genuinely ask questions about practices within programs because she had noindependent knowledge of them. The focus group leader structured group discussions aroundvariations in practices around developing social relationships. As trust developed betweenthe leader and group members, differing and even contradictory practices were articulatedby teachers from the different programs.

The feedback component of the intervention occurred in the early winter. Before theholidays, researchers independent of the intervention conducted baseline classroomobservations and ratings of classroom emotional climate, teacher responsive involvement,teacher-child attachment security, and peer social competence (described below). Weprepared a series of power-point slides and handouts with descriptive information for theseconstructs, grouped at the program, not the teacher or classroom level. The template forthese this material was standard across programs.

We used these materials in a feedback discussion with the entire teaching staff of eachprogram. At these feedback sessions we engaged in dialogue, rather than simply presentingresults. We made every attempt to present our information in a way that validated the teachers’very hard work and real successes, but wanted also to engage in some dialogue about howtheir perceptions interacted with ours, since the data provided a different lens through whichthey could view their own work. Program staff does not often have the luxury, or the meansto specifically look at, for example, how the children in their program spent their day, thecomplexity of play, or their own levels of sensitivity or consistency. The sessions wereconducted in both English and Spanish. We had a bilingual research team member present at

151Howes.p65 11/25/03, 10:30 AM61


all sessions, but in some programs it was one of the teaching staff members who translatedour work both from English to Spanish and from academic to applied.

We gave each of the programs copies of the feedback material. The feedback materialserved as baseline data and as a basis for on-going discussions between teaching staff andresearch partners who continued their weekly visits for six months. Independent researchersreturned to the programs six months after the feedback sessions and again observed andrated using the same procedures and measures as at the baseline.

Method

Participants

Early Childhood Programs. Ten early childhood programs participated in this research.Descriptive information on the programs is presented in Table 1. The programs representedboth large and small programs, a mix of funding sources, and full- and part-day programs.

As part of the selection process the project director rated each program using the EarlyChildhood Environment Rating Scale (ECERS) (Harms & Clifford, 1980). The ECERS hasbeen widely used in child development research. ECERS items are rated on seven pointscales with a 3 indicating barely adequate quality, a 5 indicating good quality, and a 7 indicatingexcellent quality. An average item score is calculated for each classroom. Two subscalescores were calculated based on previous work with the ECERS: Materials and CaregivingTone (Howes, Phillips, & Whitebook, 1992). The Materials score indicates whether or notthe program makes provisions for hands on learning and the Caregiving Tone scale indicateswhether or not the program is individually tailored to children. Table 2 includes the averageprogram scores at the start of the partnership. Six of the programs had scores in the good toexcellent range. Note that there was considerable variance in the Materials and Caregivingsubscale scores as well as in the average item scores.

Teachers. All teaching staff in the programs participated in the participant observationcomponent of the intervention. All of the teaching staff was invited to participate in theinterview portion of the research. Ninety-five percent of the teachers (n = 28) (96% women)were interviewed and included in the analysis. Thirty-six percent of these teachers wereLatino, 40% African-American, 20% white (12% Armenian immigrants) and 4% biracial.Seventy-seven percent identified their role in the classroom as teacher as opposed to assistantor teacher director. Twenty-eight percent of the teachers had BA degrees. Most of the otherteachers had only the requisite 12 units of early childhood community college classes requiredby the State.

Children. Seventy children (50% girls) participated in this research. The children wererandomly selected from each classroom in order to have approximately 25% of the childrenat the program participate. No fewer than 4 and no more than 12 children participated fromeach program. Fifty-nine percent of the children were Latino, 27% African-American, 10%White (all Armenian immigrants) and the remainder Asian or biracial. Forty-nine percent ofLatino children, 79% of African-American children and 43% of the Armenian immigrantchildren were taught by teachers from their own ethnic backgrounds. Thirty-nine percent ofthe Latino children were taught by African-American teachers. The children averaged 50.3months (SD =10.8) at the time of observation and ranged in age from 15 to 62 months.

151Howes.p65 11/25/03, 10:30 AM62


Table 1.Descriptions of Participating Programs

Children Primary Ethnicity#, ages Funding Full/part day of Families and

Neighborhood

P1 24 children Mixed Full Day LatinoAges 2-5 Public & Tuition 7 AM – 6 PM

P2 24 children Parent Tuition Full day African-AmericanAges 2-5 6 AM – 6 PM

P3 17 children Public Full Day Latino/African-AmericanAges 1-5 & Foundation Residential Teen mothers

P4 120 children Public Full Day African-AmericanAges 2-5 7 AM – 6 PM

P5 32 children Public Half day Latino/ArmenianAge 4

P6 96 children Public Half day Latino/ArmenianAges 3-4

P7 24 children Public Full day LatinoAges 1-3 7 AM – 6 PM

P8 36 children Public Full day LatinoAges 2-5 6 AM – 6 PM

P9 16 children Public Half day Latino/African-AmericanAges 2-5

P10 40 children Public Full day Latino/ArmenianAge 4 7 AM – 6 PM

Procedures and Measures

We used five distinct procedures to collect the data we analyzed in this report: (1)participant observation; (2) naturalistic observations using a snapshot procedure; (3) globalratings; (4) clinical interviews and (5) case studies. One set of researchers completed theparticipant observation and interview procedures. A different set, blind to the informationobtained in the participant observation and interviews, completed the observations and ratings.The project director recruited the programs, was primarily responsible for selecting theprograms, visited the programs frequently during throughout the project, directly supervised

151Howes.p65 11/25/03, 10:30 AM63


the research partners who served as participant observers, and wrote the case studies. Thus,different researchers collected the data pertaining to practices and quality. These proceduresand the measures associated with each procedure are described in the following section.

Each program had at least two classrooms. Each classroom had at least four randomlyselected children and at least two teachers. A primary teacher was identified for each child.A primary teacher was defined as the lead teacher in each room, unless the observed childdirected all of his or her attachment behaviors to another teacher in the classroom. In thatcase (4% of the child participants) the primary teacher was re-identified as the teacher whowas target of the child’s behaviors. All participating children had been cared for by theirprimary teacher for at least two months prior to data collection.

Participant Observation

The participant observation procedures produced field notes that were used in creatinga case study of each program. The activities of the participant observer were describedabove as a component of the intervention. Each research partner completed field notes aftereach weekly visit. The entire research team including the research partners, the project

Table 2.

Quality of Program as the Partnership Began

Program ECERS Classroom Interaction Responsive Scale Involvement

Materials Caregiving Total Sensitive Harsh Detach

P1 5.20 5.84 5.66 3.75 1.44 1.25 .54

P2 3.95 4.34 4.05 3.20 2.33 2.25 .52

P3 4.70 5.13 4.88 3.48 1.14 1.00 .42

P4 4.12 4.32 4.29 4.43 2.22 2.50 .49

P5 5.10 6.10 5.77 3.81 1.22 1.16 .36

P6 5.65 6.03 5.68 3.40 1.33 1.27 .28

P7 4.30 4.27 4.33 3.25 1.83 1.25 .49

P8 5.00 5.13 5.11 3.31 1.63 1.22 .26

P9 4.95 5.60 5.23 3.00 1.89 1.00 .27

P10 4.80 5.30 5.24 3.50 1.78 1.25 .33

Note: We used the Adult Involvement Scale to calculate the mean proportion of “responsive”involvement with the child given the total number of observations when the teacher waswithin three feet of the child.

151Howes.p65 11/25/03, 10:30 AM64


directors, the focus group leader, and the principal investigator discussed ongoingunderstandings of the programs in weekly meetings. The project director used the fieldnotes by the participant observers and transcripts of these research team meetings to createa case history of each program.

Naturalistic observations

Naturalistic observations were used to create baseline and final data on classroomemotional climate, teacher responsive involvement, and children’s peer social competence.A researcher different from the participant observer assigned to the program as a researchpartner, not necessarily of similar ethnic background, but speaking the same primary language(Spanish or English), was assigned to collect observational data from each program.

Snapshot. We used a time sampling procedure to capture the randomly selected children’sinteractions with teachers and peers. Over the course of a day (between five to eight hoursdepending on the program), 50 snapshot observations (pre- and post) were collected foreach randomly selected child. To complete a snapshot, the observer located a selected childand recorded, rating where necessary, his/her social contacts and activities. The observerthen moved on to the next child on the list. This process was repeated in five-minute blocksof time. Care was taken to space the snapshots across the course of the daily program inorder to capture the children’s experiences in the program. Snapshot data were collectedtwice, within two months of beginning the research partnership and at the end of theintervention. The same child was observed at each time period. If, as was the case for 6children, the child observed at time one was no longer in the program at the post time period,a child matched in gender, age, ethnicity, and time in program was substituted.

Eight observers were trained to an 85 percent exact agreement criterion on each snapshotitem prior to data collection. Because total observations took longer than a month to completedue to differences in class, program, and researcher’s schedules, monthly inter-observerreliability checks were conducted. Kappa statistics for agreements and disagreements ofexact snapshot interval coding are presented with the measures used in this analysis.

The snapshot content includes children’s activities, teacher behaviors with the child,and the children’s behaviors with peers. For the purposes of the analyses in this paper wewere interested in codes used for classroom emotional climate, teacher responsiveinvolvement, and social competence with peers. Embedded within the snapshot was theAdult Involvement Scale used to create the measure of teacher responsive involvement andthe Peer Play Scale used to create the measure of social competence with peers. Teachernegative management, used to create the classroom emotional climate measure, was alsocoded as an item within the snapshot procedure.

Responsive involvement. The Adult Involvement Scale (Howes & Stewart, 1987) is arating of the intensity of caregiver-child involvement. We gave the child a rating on thisscale each time the caregiver was within three feet of the child or when there was caregiver-child interaction occurring over a distance of more than three feet. Therefore, a child couldhave a maximum of 50 responsive involvement ratings each time the snapshot procedurewas used (one for each snapshot). Only one scale point could be recorded in each snapshot.

151Howes.p65 11/25/03, 10:30 AM65


The five scale points for the adult involvement codes represent increasing complexityand reciprocity in adult-child interaction. The first scale point is “monitor.” “Monitor” iscoded if the caregiver is close to the child but does not engage in interaction. “Routine,” thesecond scale point, is coded when the caregiver touches the child for changing or otherroutine caregiving but makes no verbal response to the child. Routine is also coded duringwhole group activities when the adult merely gives instructions, passes out materials, readsa book straight through without making any attempts to interact with a group of children orkeep them engaged (e.g., adult doesn’t ask clarifying or expanding questions, doesn’t pauseto answer children’s questions, etc.). “Minimal” is coded if the caregiver touches the childonly for necessary discipline, moves one child away from another, answers direct requestsfor help, or gives verbal directives with no reply encouraged. Minimal is also coded duringwhole group activities, when an adult verbally responds with one or two words (e.g., “okay,”“that’s right,” “good,”) or with mere nodding of head. In “Simple” responsive involvement,the caregiver uses some warm or helpful physical contact (beyond the essential routine care)or verbally answers the child’s verbal bids but does not elaborate. During whole groupactivities, the adult might respond to child/children with short sentences or initiate simple-social interaction (e.g., adult says, “Yes, you need to glue that piece.” “You’re doing such agood job listening!”). In “Elaborated” involvement, the caregiver engages in some physicalgestures, maintains a close proximity to the child, acknowledges the child’s statements andresponds, but does not restate, sits with the child during play, suggests materials, etc.“Elaborated” is also coded if the caregiver engaged with the group, is asking and answeringcomplex questions, acknowledges children’s statements but does not restate them, is solicitingactive participation by all group members including, but not exclusively, the target child.The highest scale point is “Intense.” This is coded when the caregiver hugs or holds thechild, restates the child’s statement, thus acknowledging them and providing answers to thechild, engages the child in conversation, plays interactively with the child, or sits and eatswith the child in a social atmosphere. Or, in group situations, “Intense” is coded if the caregiveris physically responsive to many members of the group, restates children’s ideas, engages ina conversation meant to include all group members including, but not exclusively, the targetchild. Only one rating is coded in each snapshot interval. Kappa reliabilities averaged .87(range = .79 to .83). “Responsive” teacher involvement represents the sum of the proportionof simple, elaborated and intense involvement divided by the total units of involvement.

Peer social competence. We used the Revised Peer Play Scale (Howes & Matheson,1992) to measure the complexity of peer play. The scale has eight scale points. The first fourscale points measure (1) solitary play and low level peer play: (2) onlooker behavior, (3)proximity to peer without interaction, and (4) parallel play when the target child and a peerare within three feet of each other and engage in the same activity but do not acknowledgeeach other. The other scale points capture interactive peer play: (5) parallel play with eyecontact; simple social play; (6) complementary and reciprocal play; and two levels of pretendplay: (7) cooperative social pretend (the target child and another child enacting complementaryroles), and (8) complex social pretend play (the target child and a peer demonstrating bothsocial pretend play and meta-communication about the play). Complementary-reciprocalsocial play, cooperative pretend play, and complex pretend play are considered competentforms of play with peers(Howes & Matheson, 1992). The peer play scale points were mutuallyexclusive and the highest possible observed scale point was recorded in each interval. Kappareliabilities averaged .93 (range = .86 to .95). From the coded peer play we created a measure

151Howes.p65 11/25/03, 10:30 AM66


of the proportion of the interactive play that was competent play by summing time incomplementary and reciprocal play, cooperative and complex pretend play and dividing bythe total number of units in peer play (parallel and above).

Classroom emotional climate. We coded, in each snapshot, if the teacher engaged inmanaging the child’s behavior. For this recording, the teacher was not required to be withinthree feet of the child. “Negative management” was coded if the caregiver handles the child’sproblem or misbehavior in a harsh or negative way. Examples include putting the child intime-out, yelling, criticizing, scolding, threatening, and employing sarcasm, or physical abuse.Kappa reliabilities averaged .85 (range = .80 to .92).

Global Ratings

The same observers used in the naturalistic observations spent at least four hours watchingeach randomly selected child and their primary teacher. Following the observations theobservers completed ratings on the Attachment Q-Set (Waters, 1990). Once all randomlyselected children in a classroom had been observed the observer made ratings of the teacheron the Classroom Interaction Scale (Arnett, 1989). Therefore, at each of the two time points,one Attachment Q-Set was completed for each child, and one Classroom Interaction Scalewas completed for each classroom.

Attachment Q-Set. The Attachment Q-Set (AQS) (Waters, 1990) is a measure ofchildren’s security in their attachment relationship with an adult. The AQS has good validitywith the Strange Situation (Howes & Hamilton, 1992) Following their observations, theobservers took the 90 behavior descriptions that serve as items in the AQS (for example,‘this child turns to the caregiver when she is upset’), and sorted them into nine piles of tenitems each. In this way, the observers described the teacher-child relationship according to90 behavior descriptions. Each particular behavior was given a nine-point score correspondingfrom very characteristic (behaviors in pile #9) to very uncharacteristic (behaviors in pile#1). The item scores for each teacher-child relationship were correlated with an ideal child-caregiver relationship (Waters, 1990). These became the children’s security scores. Scorescan vary from -1.0 to 1.0. A higher score indicates greater security.

Observers were trained to an 85 percent exact agreement criterion on each item prior todata collection by conducting observations in non-participating programs. Monthly inter-observer reliability checks were conducted throughout data collection. Median inter-observerreliability for exact item placement was Kappa = .87 (range Kappa = .79 to .96)

Classroom Interaction Scale. The Classroom Interaction Scale (Arnett, 1989) is a 26item four-point rating scale. It yields three scores: sensitivity (warm, attentive, and engaged);harshness (critical, threatens children, and punitive); and detachment (low levels of interaction,interest, and supervision). Median inter-observer reliability for exact scale point on eachitem was Kappa = .85, (range Kappa = .79 to .88). Scores from this instrument have beenfound to predict child care caregiver’s involvement with children and the children’s socialcompetence (Howes, Phillips, & Whitebook, 1992) and attachment security with teacher(Howes & Hamilton, 1992).

Clinical Interviews

In the fall we conducted individual interviews with all program directors and teachingstaff who wished to be interviewed. No program directors declined and 95% of teachingstaff were interviewed. The remaining 5% declined an interview due to time conflicts.

151Howes.p65 11/25/03, 10:30 AM67


Interviews were conducted in either English or Spanish, depending on the preference of theparticipant. The interviews were structured like a conversation, and questions fell under thefollowing general headings: What do you do in the program to help children learn and developand why?; How did you become a teacher?; How would you describe the families and childrenwho use the program?

For our analysis, each interview was transcribed, translated if applicable, and coded foremerging patterns. Interview transcripts were repeatedly read and discussed by the researchteam. Content codes were derived by repeated reading of the interviews, consensus agreementon codes, and coding at two levels. One set of codes was applied to each meaningful unitwithin the transcript. Meaningful units were utterances, sentences, or in some rare cases,paragraphs. Inter-coder reliability for segmentation was Kappa = .98 (range .95 to .99).Each meaningful unit was then coded for content category (e.g., teaching and learningpractices). All units with the same content category codes were then collapsed withininterviews and a most prevalent and salient content category score was given to eachparticipant. Inter-coder reliability for content category codes ranged from Kappa = .86 toKappa = .95.

In this analysis we were interested in the content categories of teacher articulated socialpractice. Each teacher received a mutually exclusive score. The three categorical scoreswere: (1) No mention of socialization or social competence; (2) Behave in school (Teachthe children how to behave in school); and (3) Work on social relationships (teaching;nonviolent conflict resolution skills; talking about dramatic play as an important social activityfor children to develop peer relationships; and/or help children learn to work and play together,and to become socially competent). Further elaboration of these practice codes can be foundin Wishard, et al., (2003). Twenty-five percent of the children had teachers who did notmention socialization, 26% who talked about behaving in school, and the remaining 49%spoke of promoting social relationships. Formal education codes also were created from theinterview. For the purposes of this analysis teachers were classified as having or not havinga BA in a child related field.

Case Studies

We used the individual program case studies to code three measures of teacherengagement with the partnership project. We coded whether or not the teacher was part ofthe focus group activities, whether or not she participated in the feedback sessions, and arating of positive engagement with the program. Focus group participation codes were “1”never participated (either by not being selected as a teacher representative or by being selectedbut failing to attend), “2” attended one or two meetings (selected as a teacher representativebut intermittent participation), and “3” attended all meetings (selected as a teacherrepresentative and a faithful attendee). Feedback sessions codes were “1” did not attend, “2”attended but did not participate or did not discuss feedback session with research partner,and “3” attended and either participated or discussed content with research partner. Positiveengagement codes were “1” merely tolerated or actively resisted the intervention process,“2” Enjoyed the activities and the research partner as an extra pair of hands in the classroombut did not engage with the ideas, or “3” fully engaged with the research partner in makingsense of the ideas and interactions of the projects. To receive a positive rating on positiveengagement a teacher did not have to agree with suggested changes in practices, but they

151Howes.p65 11/25/03, 10:30 AM68


could not be disengaged or actively working against the process of change. The principalinvestigator made these rating for each teacher participant using the program case studies.

Constructs Used in Analysis

Outcome Measures

Children attachment to teacher. We used the Attachment Q-Set security scores torepresent teacher-child relationship quality. At the initial data collection children’s securityscores averaged .23, and ranged from -.37 to .69.

Children’s social competence with peers. The proportion of the time the child wasinvolved in competent peer play represented children’s social competence with peers. Theproportion of time spent in competent peer play averaged .10, and ranged from 0 to .30.

Social and emotional climate

Classroom interaction scale scores at time 1, averaged to the program level are in Table2. These classroom level scores range from relatively low levels of detachment and harshnessto moderate levels. The modal participating child experienced no negative management,but a small proportion of the children experienced negative management as often as 17% ofthe observation period.

To create a classroom social and emotional climate score for each child, we first convertedClassroom Interaction Scale scores and the proportion the observation period experiencingnegative management from the snapshot into standard scores. We then subtracted the harshnessand detachment standard scores and the standardized proportion of negative managementfrom the sensitivity standard score.

Responsive involvement.

We used the Adult Involvement Scale to calculate the mean proportion of “responsive”involvement with the child given the total number of observations when the teacher waswithin three feet of the child. Average program scores at time 1 also are in Table 2.

Results

Engagement with the Intervention

We began our analysis by examining the implementation of the intervention using thescores for teacher engagement with the intervention. Recall that each participating teacherhas three scores for engagement (with the focus group, with the feedback sessions, and ingeneral with the intervention). We gave each child the scores for their primary teacher andconducted the analysis with child as the unit of analysis.

Only one-third of the children had teachers who participated fully in the focus groups.(See Table 3). This was in part because two programs for logistic reasons did not participatein the focus groups, and because for the remaining eight programs only two teachers fromeach program participated. Only a few (6%) children had teachers who were selected toparticipate in the focus groups and did so only sporadically.

151Howes.p65 11/25/03, 10:30 AM69


In contrast to the focus groups, all children had primary teachers that had an opportunityto participate in feedback sessions. Less than one-quarter of the children were cared for bya teacher who was given low ratings for participation in the feedback sessions. It is importantto note that over half of the children were cared for by teachers given high ratings onparticipation in feedback sessions. See Table 3.

Table 3.

Engagement with the Intervention

Focus group Feedback Process Participation as a Whole

Low 61% 23% 36% 40%

Medium 6% 25% 23% 18%

High 33% 52% 41% 42%

Note: Numbers in table are proportion of children

Somewhat fewer than half of the children (42%, see Table 3) were cared for by primaryteachers given high ratings for participation in the intervention process as a whole. Andapproximately one-third of the children were cared for by primary teachers with low ratingswho merely tolerated or resisted the process. These three teacher engagement variables(participation in focus groups, participation in feedback, participation in the process as awhole) were highly intercorrelated (r’s range .34 -.87). Therefore we created a single summedstandardized variable for primary teacher engagement.

Changes in children’s social competence, social-emotional climate, and responsiveinvolvement from baseline to the end of the intervention.

We began our analysis of change by exploring simple changes over time. To do this weused a series of multivariate repeated measures analyses of covariance comparing behaviorsat times 1 and 2, and using child’s age as a covariate. The descriptive statistics, F values, andeffect sizes are shown in Table 4.

There were significant changes in children’s social competence. Children’s securityscores and their complex play with peers increased over time. Responsive involvement wassignificantly higher after the intervention than before. Social and emotional climate did notdiffer statistically after the intervention as compared to before. It is important to note that themagnitude of these variables at times 1 or 2, and the magnitude of the change in these values,was not correlated with children’s age, or different for girls and boys.

Understanding change

To quantify change we created change scores for social competence (peers and security),climate, and responsive involvement by subtracting time 1 scores from time 2 scores. As canbe seen in columns 5 and 6 in Table 4 the change scores for all outcomes ranged fromnegative, indicating that the behavior became less positive, to positive indicating that positive

151Howes.p65 11/25/03, 10:30 AM70


Tabl

e 4.

Cha

nges

in C

hild

ren’

s So

cial

Com

pete

nce,

Soc

ial-e

mot

iona

l Clim

ate,

and

Res

pons

ive

Invo

lvem

ent F

ollo

win

g In

terv

entio

n

T

ime

1

Tim

e 2

Ran

ge o

f Cha

nge

F c

hang

e

Eta

2

MSD

MSD

Lo

w

Hig

h

Soci

al C

ompe

tenc

e4.

23*

.12

Secu

rity

.23

.26

.29

.23

-.70

.99

3.64

*.0

7

Peer

.10

.09

.14

.11

-.24

.43

5.86

**.0

9

Soci

al-e

mot

iona

l Clim

ate

.00

3.56

.00

2.35

-5.6

29.

14.0

4.0

0

Res

pons

ive

Invo

lvem

ent

.37

.02

.45

.03

-.64

.80

4.80

*.0

7

*p <

.05;

**p

< .0

1

151Howes.p65 11/25/03, 10:30 AM71


change occurred. This indicates that there is considerable variation in individual children’sexperiences of change and sets the stage for an analysis of which children actually experiencedpositive changes over the course of the intervention.

Change scores were not significantly intercorrelated (change in security and change inpeer, r (68) = .15, ns; change in security and change in climate, r (68) = .21, ns; change insecurity and change in involvement, r (68) = .12, ns; change in peer and change in climate,r (68) = .22, ns; change in peer and change in involvement, r (68) = .13, ns; change inclimate and change in involvement, r (68) = .03, ns).

Recall that we expected that teacher education level, engagement with the interventionprocess, and teacher articulated practices around social behavior would predict which childrenexperienced change. We suspected that the predictor variables were not independent of eachother. Therefore, we first examined associations among these predictors. In further examiningindividual differences in change we examined associations between individual predictorsand change values holding constant intercorrelated predictors. Finally, we use a hierarchicalmultiple regression strategies to examine relations between predictors and change scores.

Associations among predictors

As expected, teachers with BA degrees had higher engagement with the interventionscores than teachers without BA degrees (t (27) = 2.53, p < .05). Teachers with BA degreeswere more likely than teachers without BA degrees to articulate a practice around socialbehavior (c2 (1) =14.50, p. = .001) but no more likely to articulate one social practice thanthe other (c2 (1) =.58, ns). Teachers who articulated the practice of teaching children tobehave in school were less engaged in the intervention than teachers who either did notmention social behavior or articulated practices around social relationships (F (1,67) = 8.01,p < .001, eta2 = .12, Scheffe = .05).

Associations between individual predictors and change

BA degrees. We compared differences between children, with teachers with BA degreesand no BA degrees, in the children’s experiences of change in social competence, climate,and responsive involvement. We used a series of multivariate analysis of co-variance, holdingconstant social behavior, articulated practice, and engagement scores. Children with teacherswith BA degrees had greater changes in climate than children with teachers without BAdegrees (F (1,66) = 4.56, p < .001, eta2 = .07). There were no significant differences forchanges in social competence or responsive involvement.

Engagement. We used partial correlations, holding BA and articulated practices constant,to examine associations between engagement scores and change scores. All of the associationswere significant. Children with higher teacher engagement scores also experienced greaterchanges in climate change scores (r [66] = .28, p < .05), responsive involvement changescores (r [66] = .27, p < .05), complex peer play change scores (r [66] = .36, p < .05), andsecurity of attachment change scores (r [66] = .28, p < .05).

Articulated social practices scores. We compared differences between children withteachers with different articulated social practice scores, in the children’s experience ofchange in social competence, climate, and responsive involvement. We used a series of

151Howes.p65 11/25/03, 10:30 AM72


multivariate analysis of co-variance, holding constant formal education and engagementscores. There were significant main effects for social competence (Multivariate F [4,118] =4.10, p < .04, eta 2 = .12, Security scores: F [2,63] = 4.17, p < .02, eta2 = .12, Scheffe = .05;Peers: F [2,63] = 4.83, p < .01, eta2 = .14, Scheffe = .05), social emotional climate (F [2,65]= 4.75, p < .01, eta 2 = .13, Scheffe = .05) and responsive involvement (F [2,65] = 5.45, p <.01, eta2 = .16, Scheffe = .05). Post hoc Scheffe tests (p < .05) indicated that children withteachers who initially articulated practices around social relationships experienced morechange in social competence with peers than children with teachers who did not articulatesocial practices or articulated social practices around behavior. Children with teachers whoinitially articulated social practices (behavior or relationships) had greater increases in theirsocial and emotional climate scores than children with teachers who did not articulate socialpractices. Finally, children with teachers who initially indicated practices around havingchildren behave well when they go to school had greater increases in their teacher responsiveinvolvement scores than children with teachers who initially did not articulate a social practiceor articulated a social practice around relationships.

Predicting change

Based on our above analyses of, among, and between predictors and change we devisedthe following model to use in a series of hierarchical multiple regressions to predict changes.We entered as a first block formal education, engagement and articulated social practicesdummy coded as “behave in school” and “social relationships.” Because engagement wasso highly correlated with the formal education and with initially articulated social practiceswe entered as a second block the interaction of engagement and formal education, and theinteraction of engagement and articulated social practices.

As expected, given the preceding analyses, for all four of our outcome variables changescores could be significantly predicted by the first block of predictors: formal education,engagement, and articulated practice (security: R = .40, p = .01; peers: R = .40, p = .01;climate: R = .41, p = .01; and responsive involvement: R = .40, p = .01). The second blockof predictors added significantly to the predictive power of the regression only for changesin responsive involvement and changes in social competence with peers. In both analysesthe addition of the interaction terms resulted in a .07 change in R2. In predicting change inresponsive involvement there was a significant b for the interaction between engagementand articulated practice. Figure 1 describes this interaction. Children cared for by teacherswho were engaged with the intervention and did not initially articulate practices aroundchildren’s social behavior were most likely to have an increase in teacher responsiveinvolvement.

There was a significant b for the interaction between engagement and formal educationin predicting change in social competence with peers. This interaction is shown in Figure 2.Children whose teachers did not have BA degrees and were engaged in the intervention hada greater increase in social competence than those children who had teachers without BAdegrees who did not engage in the intervention. This second group of children decreasedsomewhat in social competence.

151Howes.p65 11/25/03, 10:30 AM73


0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

Time 1 Time 2

No Engage and No BA

Engage and No BA

Engage and BA

Figure 2.

Social competence with peers.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

Time 1 Time 2

Engaged and No PracticeNot Engaged and BehaveNot Engaged and RelationshipEngaged and Relationship

Figure 1.

Teacher responsive involvement.

Discussion

The relationship-based intervention described in this report was overall associated withpositive changes in child-teacher attachment relationships, the responsiveness of teachers tochildren, and in the complexity of children’s play with peers. These changes were not simplyrelated to changes in children’s ages, but appear to be rooted within variations in teacherengagement with the intervention. These findings suggest a potentially promising directionfor intervention with early childhood teachers. However, these simple main effects of theintervention need to be qualified in several substantial ways. This study was carried out

151Howes.p65 11/25/03, 10:30 AM74


without a control group or random assignment to intervention, essential for establishing thatthe intervention is causally related to the outcomes. Instead this report is a description of anintervention process and will require more rigorous evaluation.

It is possible that there may be an inherent contradiction between relationship-basedintervention and true experimental design. If we had randomly selected the participants andassigned them to control or intervention, it would have been harder to begin our work withthe centers based on the assumption that we were entering into a relationship together toinstitute change. In this light it is important to understand the particular nature of the sampleof child care centers selected to participate in the intervention. At least one person in theparticipating programs was initially motivated to engage in a change process. And in makingour selections we only included programs in which the teaching staff eagerly agreed toparticipate. We made sure that the staff understood the demands (and we hoped possibleadvantages) of participation by having extensive conversations between the project directorthe research partner and the staff at the beginning of the intervention.

We learned that initial full disclosure and working to cooperate with the teaching staffrequired that we act on our principles of relationship building in our construction of ourresearch design as well as in our intervention. This too would have been more problematic ifwe had been engaged in a true experiment. Relationships are constructed with the participationof both partners. This meant that as we negotiated our participation as researchers with theprograms that we, at times, did something somewhat different than we had first designed.For example, in one program, the lead teacher asked that the participant observer researchpartner not be involved in a fairly elaborated naptime practice. Each child had anindividualized script for sleep, and the entire naptime routine took at least 45 minutes andsometimes as long as an hour or more. The research partner was asked to leave the classroomduring this time. After seven months of weekly visits, the teacher asked that the researchpartner take her turn in the nap room, including her with the rest of the staff. In anotherprogram the first author had planned to be an additional participant observer, making weeklyvisits along with the research partner. The staff asked that she not participate, worrying thatthe age and experience discrepancies were too great. We agreed with this request and thefirst author did not engage with the program. Finally, in another program the staff was initiallysplit over their participation in the process. The program director and several teachers beggedto be included. Other teachers in the program “went along” and one of them became one ofthe representatives to the focus group. However, this teacher was only a sporadic attendee atthe focus groups and when present was disruptive. As researchers, we could not ask the staffto send a different representative and this program had the largest proportion of not engagedteachers of all of the programs.

Another major limitation on our ability to generalize this work is the relatively highresource level of both the programs and the research. All of the programs had relatively highenvironmental quality scores. The participating programs were not struggling with too fewresources to provide experiences that were safe for children. The intervention and researchitself also were very well funded. The programs may have been particularly responsive tothis intervention and the researchers data collection effects because they were not paying forit and did not have to make the hard choices between paying for a quality improvement orpaying for services. If we had conducted the intervention in the current era of communityinitiated quality improvements we suspect it would have been more difficult.

151Howes.p65 11/25/03, 10:30 AM75


Another potential limitation on generalizing these results is that the teachers and thechildren in the programs were predominantly Latino and African-American, and that programstended to have children and teachers from one ethnic group rather than multi-racial groups.We suspect that practices rooted within these ethnic communities interacted with theintervention (Wishard et al., 2003).

Not all of the teachers engaged in the intervention. Even when the program had made acommitment to the intervention, fully one-third of the teachers merely tolerated or activelyresisted the intervention process. Teacher engagement was important to understanding theprocesses under what circumstances children’s experiences in the programs changed. Wesuspect that the teachers who participated more fully in the intervention process could havedone so due to their pre-existing skills, knowledge, or dispositions. Certainly completion ofa BA degree in a child related field is a marker for skills and knowledge and probably, in thissample, for the dispositions and freedom from devastating life circumstances that allowfinishing a degree while teaching in child care. And of course, teachers were more likely toengage in the intervention process if they had BA degrees. Perhaps these more highly educatedteachers found it easier than the less educated teachers to construct a relationship with theresearch partners because they were less constrained by social status differences, or morereceptive to knowledge obtained from research as well as or instead of the lived experiencesof mothering. Alternatively, as others have speculated, a BA degree permits the teacher tomore easily reflect on the process of teaching (Howes, James, & Ritchie, 2003), as theintervention demanded. We found as well that teachers with BA degrees were more likely toarticulate teaching practices around relationships than around having children behave inschool, suggesting that they were receptive to ideas around building relationship as theintervention began.

While the intervention was most accessible to teachers who had pre-existing skills,knowledge and dispositions as marked by the BA degree, some teachers without BA degreesengaged with the intervention. When we held BA degrees constant, children who had engagedteachers experienced the greatest changes in classroom climate, teacher responsivity, and intheir relationships with the teacher. These findings suggest that the intervention was effectivefor teachers with and without high levels of formal education in a child related field. Furtherwork is needed to understand whether the strategies for engaging non-BA teachers inrelationship-based intervention are the same or different than the strategies for engaging BAlevel teachers. As well future research needs to examine the efficacy of various strategies forincreasing engagement in teachers with different belief systems and backgrounds.

Although not predictive by themselves, teacher-articulated practices at the beginning ofthe intervention did interact with engagement and BA degrees to explain change. In ouranalysis changes in responsive involvement were greatest for those children with teacherswho did not articulate a social practice and did engage in the intervention. In prior workwith this sample we did find that teacher-articulated practices were associated with teachers’classroom activities (Wishard, et al., 2003). Our participant observer experiences areconsistent. For our research partners the lack of practices within classrooms often presenteda challenge because the teachers who were unable to articulate their intended practices aroundsocial relationships and social behavior struggled with classroom management. The followingexample comes from the field notes of a research partner of such a teacher.

151Howes.p65 11/25/03, 10:30 AM76


Second time at this site, first time in Ms. Q’s class: Most notable observance is howinappropriate Ms. Q is with the children and how the other teachers don’t really intervenealthough they make subtle attempts. She is easily irritated with the children, makes negativecomments to them when they don’t do something the way she wants. However, when sheexplains rules to game she is not very clear, it is very confusing…it borders on threateningthe way she talks to the children.

Ms. Q and this research partner did form, over time, a very positive relationship andMs. Q was one of the teachers whose behavior with children changed. She later told theresearch partner how frustrated she had been with her self-perceived lack of teaching skillsand that getting to know the research partner had helped her understand why her approachdid not work and that she needed to listen more closely to what the children were saying. Wedo not know what would have happened if we had taken a more skill-based approach withMs. Q, but our findings suggest that the relationship-based intervention helped her botharticulate a different practice and act on it.

There are important implications of this research for improving the experiences ofchildren in early childhood programs. The intervention, based in theory, may have beeneffective in changing in the short term, relationships and behaviors in the classroom. It washowever an intensive and fairly expensive intervention involving very well educated andsupervised interveners who spent a great deal of time in the classrooms that were not themost typical classrooms in early childhood programs. Further research is needed to determineif such an approach would be as effective if the intervention occurred in programs withdifferent resources, within different ethnic communities, or with less educated and/orsupervised interveners.

References

Arnett, J. (1989). Caregivers in daycare centers: Does training matter? Applied DevelopmentalPsychology, 10, 514-552.

Denham, S. A., & Burton, R. A social-emotional intervention for at risk 4-year-olds. Journal ofSchool Psychology, 34, 225-245.

Erickson, M. F., & Kurz-Riemer, K. (1999). Infants, toddlers, and families: A framework forsupport and intervention. New York: Gilford.

Harms, T., & Clifford, R. M. (1980). Early childhood environmental rating scale. New York:Teacher’s College Press.

Honig, A. S., & Hirallal, S. (1998) Which counts more for excellence in child care staff: yearsin service, education level, or ECE course work? Early Child Development and Care,145, 31-46.

Howes, C. (2000). Social-emotional classroom climate in child care, child-teacher relation-ships, and children’s second grade peer relations. Social Development, 9, 191-204.

Howes, C., & Hamilton, C. E. (1992). Children’s relationships with child care teachers: Stabil-ity and concordance with maternal attachments. Child Development, 63, 879-892.

151Howes.p65 11/25/03, 10:30 AM77


Howes, C., James, J., & Ritchie, S. (2003). Pathways to effective teaching. Early ChildhoodResearch Quarterly, 18, 104-120.

Howes, C., & Matheson, C. C. (1992). Sequences in the development of competent play withpeers: Social and social pretend play. Developmental Psychology, 28, 961-974.

Howes, C., Phillips, D. A., & Whitebook, M. (1992). Thresholds of quality in child care centersand children’s social and emotional development. Child Development, 63, 449-460.

Howes, C., & Ritchie , S. (2002). A matter of trust: Connecting teachers and learners in theearly childhood classroom. New York: Teacher’s College Press.

Howes, C., & Stewart, P. (1987). Child’s play with adults, toys, and peers: An examination offamily and child-care influences. Developmental Psychology, 23, 423-430.

Lieberman, A., & Zeanah, C. H. (1999). Contributions of attachment theory to infant-parentpsychotherapy and other interventions with infants and young children. In J. Cassidy& P. R. Shaver (Eds.), Handbook of Attachment Theory and Research (pp. 555 - 574).New York: Gilford.

Rogoff, B. (2002). The cultural nature of human development. New York: Oxford UniversityPress.

Rubin, K. H., Bulowski, W., & Parker, J. (1998). Peer interaction relationships and groups. InN. Eisenberg (Ed.), Handbook of Child Psychology; Vol. 3, Social, Emotional andPersonality Development (5th ed.). New York: John Wiley & Sons.

Shonkoff, J. P., & Phillips, D. A. (Eds.) (2000). From Neurons to Neighborhoods. Washington,DC: National Academy Press.

Sroufe, L. A. (1983). Infant-caregiving attachment and patterns of maladaptation in preschool:The roots of maladaptation and competence. In M. Permutter (Ed.), Minnesota sym-posium on child psychology (Vol. 16, pp. 41-81). Hillsdale, NJ: Erlbaum.

Van Ijzendoorn, M. H., Juffer, F., & Duyvesteyn, M. (1995). Breaking the intergenerationalcycle of insecure attachments: A review of attachment-based interventions on mater-nal sensitivity and infant security. Journal of Child Psychology and Psychiatry, 36,225-248.

Waters, E. (1990). Appendix A: The attachment q-set (version 3.0), Monographs of the Societyfor Research on Child Development (Vol. 60, pp. 234-246).

Wishard, A., Shivers, E., Howes, C., & Ritchie, S. (2003). Child care programs and teacherpractices: Associations with quality and children’s experiences. Early Childhood Re-search Quarterly, 18, 65-103.

151Howes.p65 11/25/03, 10:30 AM78

Improving Social Relationships in Child Care through a Researcher-Program Partnership

Documents

Transcript of Improving Social Relationships in Child Care through a Researcher-Program Partnership